A Numerical Study of Transport Phenomena in Porous Media
by
May-Fun Liou
Submitted in partial fulfillment of the requirements For the degree of Doctor of Philosophy
Thesis Adviser: Dr. Isaac Greber
Department of Mechanical and Aerospace Engineering
Case Western Reserve University
August, 2005
CASE WESTERN RESERVE UNIVERSITY
SCHOOL OF GRADUATE STUDIES
We hereby approve the dissertation of
______
candidate for the Ph.D. degree *.
(signed)______(chair of the committee)
______
______
______
______
______
(date) ______
*We also certify that written approval has been obtained for any proprietary material contained therein.
For my beloved parents,
Bow-Churng
&
Chung Wan
ii Table of Contents
List of Tables vii
List of Figures viii
Nomenclature xvi
Abstract xx
1 Introduction 1
1.1 General Overview 2
1.2 Volume Averaging Method 8
1.3 Three Dimensional Navier-Stokes Equations 13
1.3.1 Fluid-phase Equations 13
1.3.2 Solid-phase Equation 15
1.4 Material under Studied 15
1.5 Outline of the Thesis 16
2 An Enabling Technique for Pore-scale Modeling 18
2.1 Introduction 18
2.2 Mesh-based Microstructure Representing Algorithm 22
2.2.1 Mesh Generator 28
2.2.2 Random Number Generator 32
2.2.3 Quality Control 34
2.2.4 Porosity 37
2.3 Formation of the Two - dimensional Geometry 38
iii 2.4 Formation of the Three - dimensional Geometry 40
2.5 Numerical Solutions 43
2.5.1 Governing equations - both fluid and solid phases 43
2.5.2 Boundary conditions 44
2.5.3 Numerical methods 46
2.6 Sensitivity Test 47
2.6.1 Geometric types 47
2.6.2 Grid dependence study 47
2.6.3 Convergence study in terms of residual 49
2.7 Discussion 50
2.8 Summary 51
3 Fluid Flow and Pressure Drop in Porous Media 53
3.1 Introduction 53
3.2 Validation Cases 56
3.2.1 Laminar flow over a square cylinder 56
3.2.2 Laminar flow over a circular cylinder 57
3.3 Numerical Experiments of 2D Porous Flow 60
3.4 Results and Discussion 61
3.4.1 Two-dimensional channel flow 62
3.4.1.1 Velocity field 63
3.4.1.2 Pressure drop and porosity 70
3.5 Summary 73
iv
4 Forced Convection Heat Transfer in Porous Media 74
4.1 Introduction 74
4.2 Validation Cases – forced convection in single plate-plate channel 84
4.2.1 Bottom plate as an inner plate 86
4.2.2 Bottom plate as an external plate 90
4.3 Numerical Setup for Forced Convective Heat Transfer 94
4.3.1 Two-dimensional channel flow 95
4.3.1.1 Results and discussion 97
4.3.1.2 Effective thermal conductivity 117
4.3.1.3 Thermal dispersion 122
4.3.2 Summary 137
4.4 Three-dimensional Rectangular Duct Flow 138
4.4.1 Flow with Heated Porous Section 138
4.4.2 Summary 150
4.5 Conclusion 150
5 Closure
5.1 Summary 152
5.2 Future Work 154
Appendix A Programs of Random number generating 156
v Appendix B Governing equations and numerical methods 160
Governing equations 160
Numerical method 162
Appendix C Fluid flow and pressure drop data 168
References 169
vi List of Tables
Table 1 : List of thermophysical properties of solid materials under study……… 16
Table 2 : Results from grid-dependence study………………………………….. 48
Table 3 : Summary of previous works in evaluating viscous resistance (pressure drop) in porous media……………………………………………………….. 54
Table 4 : List of pressure drops for various porous systems…………………….. 61
Table 5 : Summary of numerical simulations of forced convective heat transfer in Porous media………………………………………………………….. 80
Table 6 : The effective heat conductivity of fluid phase at Re = 270…………… 120
Table 7 : The ratio of effective heat conductivity of fluid phase, f, at Re = 1,100. 121
Table 8 : The ratio of effective heat conductivity of fluid phase, f, at Re = 2,200.. 122
Table C-1: The pressure drop of a simplified electronic cooling system at various conditions………………………………………………………………. 168
vii List of Figures
1.1 Schematic of porous medium, flow variations in the representative Elementary
Volume and the pore velocity vector Vp...... 8
2.1 A Schematic explaining the procedures of generating a numerical porous
sample………………………...... 28
2.2 A porous sample with non-uniform porosity……………………………… 34
2.3 Schematic of typical domain of a porous medium – two phase problem…. 37
2.4 Randomly formed solid matrix representing porous medium in 2-D domain
(a) fibrous shape formed with triangular mesh,
(b) small solids in circular shape formed with triangular mesh,
(c) large solids in circular shape with hybrid mesh……………………… 39
2.5 Open-cell foam microstructure …………………………………………… 41
2.6 (a) the cross-section of a typical mesh-based porous medium is shown
with solid balls in black and space taken by fluid in white……………. 41
2.6 (b) The cross-section of numerical generated open-cellular sample after
converting all solid and fluid tags from Figure 2.8(a), into the oppose
ones……………………………………………………………………... 42
2.6 (c) 3D Numerical fluid and solid mesh in cross cut planes along the axial
direction – x direction. …………………..……………………………… 42
viii 2.7 Convergence history of residuals from solving Navier-Stokes equations for
fluid and solving heat conduction equations for solid in the case of the channel
flow heated from both top and bottom walls. …………………...………… 50
3.1 Streamlines of a low Reynolds number flow exhibits symmetry around the
solid block ………………………………………………………………… 57
3.2 A hybrid mesh is used for studying unsteady laminar flow over a stationary
circular cylinder. ………...………………………………………………… 58
3.3 Contours of u velocity indicate that the laminar flow is fully developed in the
flow field …..……………………………………………………………… 59
3.4 Schematic numerical setup for measuring pressure drop …………………... 60
3.5(a) The velocity distribution along the distance from the bottom wall, y', with
respect to sampling locations of ∆1, ∆2, ∆3, ∆4, ∆5 and at the exit of the
channel …...………………………………………………………………. 64
3.5(b) The velocity distributions in the axial (X) and vertical direction (Y)
respectively ..………… ..………………… ……………………………… 65
3.6(a) Velocity contours of the whole numerical domain for the inflow velocity
0.1m/s ………………………..……………………………………….... 66
3.6(b) Vortices contours of the whole numerical domain for the inflow velocity
0.1m/s …………………………………………………………………… 67
3.7 (a) The pressure distribution along the non-dimensional axial direction with
respect to several y locations close to the wall……………………. …… 68
3.7 (b) The pressure distribution along the non-dimensional axial direction
at three y locations, farther away from the wall………………………... 69
ix ∗ 3.8 Dimensionless pressure gradient parameter, P vs. Red, of various porous
systems in Table 3 ………………………………………………………….. 70
∗ 3.9 Non-dimensional pressure drop P vs. Red in a wide range of Reynolds
numbers for two values of porosities.……………………………………… 71
∗ 3.10 A close-up view of the non-dimensional pressure drop P vs. Red, exhibiting
different behaviors caused by varying porosityε …………………………. 72
4.1 Schematic cooling of electronic board with a substrate between coolant and
board. ………………………………………………………………………. 85
4.2 Silicon rubber substrate with isothermal walls at 900 K, (a) velocity contours,
(b) temperature contours…………………………………….……………… 87
4.3 Local Nusselt number Nux versus dimensionless x of heat transfer at the contact
surface of coolant and substrate of silicon rubber………………………. 88
4.4 Aluminum substrate with constant wall temperature, Tw=900K, (a) velocity
contours, (b) temperature contours……………………………………… 89
4.5 Silicon rubber substrate with Tw = 900 K and the other parallel plate insulated
(a) velocity contours, (b) temperature contours………………………… 90
4.6 Local Nusselt number Nux versus dimensionless x of heat transfer at the
contact surface of coolant and substrate of aluminum………………… 92
4.7 Aluminum substrate with isothermal top wall at Tw = 900 K, and the lower
parallel plate insulated: (a) velocity contours, (b) temperature contours. 93
4.8 Nux along longitudinal direction x at the contact surface of coolant and
substrates of silicon rubber and aluminum…………………………….. 94
x 4.9 Schematic of a two-dimensional porous channel with heat source from
below or above…………………………………………………………. 95
4.10(a) Temperature contours of the nonporous channel with the background
showing the mesh used………………………………………………… 99
4.10(b) The invariant temperature profiles along the transverse direction in the
channel at various sampling positions along the axial direction………. 99
4.10(c) Velocity profiles along the transverse direction at all sampling positions.100
4.10(d) Local Nusselt number Nux versus the longitudinal coordinate x of the
baseline case…………………………………………………………… 100
4.11(a) Temperature contours of the porous sample in a channel, along with the
mesh used and the porous region in which the solid matrices are shown in
black…………………………………………………………………… 101
4.11(b) The invariant temperature profiles along the direction of y* in the channel
are plotted for various sampling positions……………………………. 102
4.11(c) Profiles of non-dimensional x-component velocity along the transverse
direction at various sampling locations………………………………… 104
4.11(d) Plot of local Nusselt number Nux versus the longitudinal coordinate x… 104
4.12 The ratio, f, of local heat conductivity of fluid phase to that at the inflow
condition (300K and 1atm) in terms of the distance away from wall at
various sampling locations……………………………………………… 105
4 .13 Distributions of local dimensionless heat transfer along the bottom wall of
four solid phases, the upward spikes indicate the enhancement of heat
transfer contributed from the physical blockages of porous media……. 108
xi 4.14 Vorticity distribution along the dimensionless transverse direction, y, on the
mid-plane of the porous medium (∆3)…………………………………. 109
4.15 Vorticity distributions on the center plane of the hypothetical solid/air
(system of θ = 1.0) under Case (b)……………………………………... 111
4.16 Local dimensionless heat transfer rate along the bottom plate at various
Reynolds numbers. The porous system studied is composed of air and the
hypothetical solid/air as the solid phase, i.e. θ =1……………………… 112
4.17 (a) - (c) The local Prandtl number profiles along the distance away from the
heated wall at various x* locations, where — 0.25 (∆1), — 0.45 (∆2), and —
0.5 (∆3) — 0.55 (∆4) , and — 0.75 (∆5), of (a) Re = 1,100 (b) Re = 2,200 and
(c) Re = 5,500…………………………………………………………… 114
4.18 (a) - (e): Profiles of effective heat conductivity at five planes, ∆1-∆5,
respectively shown in (a)-(e), for three Re=1,100 (─), 5,500 (─), and 11,000
(─)………………………………………………………………………. 115
4.19 Local Nusselt number along the bottom wall, showing the effects of Re, heat
conductivity and specific heat of solid phases…………………………. 116
4.20 Profiles of ratio of heat conductivities f in the transverse direction with
respect to two inflow conditions (Re)………………………………….. 126
4.21 The spatially averaged v-velocity along the longitudinal direction is plotted
with the fluctuation, v', deviated from it at the location of ∆4, Re=5,500.
Solid symbol denotes the spatial average and open symbol denotes the
fluctuation……………………………………………………………… 128
xii 4.22 The temperature fluctuation, T' (K), is plotted together with the spatially
averaged temperature, Tavg as indicated in the figure; data are taken at ∆4.129
4.23 The component of Kd in the longitudinal coordinate varies along the height
of channel with respect to various flow rate (Reynolds number)……… 130
4.24 The component of Kd in the transverse coordinate varies along the height of
channel with respect to various flow rate (Reynolds number)………… 131
4.25 The thermal dispersion conductivity in the x and y direction versus the
Reynolds number under case (b)………………………………………. 132
4.26 The spatially averaged v-velocity along the longitudinal direction is plotted
with the fluctuation, v', deviated from it at the location of ∆4, Re=550... 133
4.27 Profiles of non-dimensional diffusivity in the transverse direction with
respect to various Péclet number, under wall condition of Case (a) –
non-symmetric heating………………………………………………….. 134
4.28 Profiles of non-dimensional diffusivity in the transverse direction with
respect to various Re, under wall condition of Case (b) – symmetric
heating…………………………………………………………………… 135
4.29 Plots of analyses under heating condition of case (a) to summarize thermal
dispersion study………………………………………………………… 137
4.30 Comparison of heat transfer effect in Nusselt number with respect to two
heating cases, case (a) and case (b)…………………………………... 138
4.31 Schematic of a three-dimensional porous duct with heated wall…….. 140
4.32 Contours of temperature and density of the porous system with meshes.. 141
4.33 Contours of density of Y-Z planes in the longitudinal direction of x… 142
xiii 4.34 The axial velocity contours at crossplanes (Y-Z) along the longitudinal
direction………………………………………………………………. 143
4.35 Detailed contours of axial velocity and temperature at the (a) mid-plane (∆3)
and (b)downstream (∆4 )of the porous sample……………………….. 144
4.36 Temperature contours in Y-Z planes along the axial direction x…….. 145
4.37 Contours of v, the velocity in the transverse direction of y at five Y-Z planes
along the axial direction x…………………………………………….. 147
4.38 Contours of w, the velocity in the transverse direction of y at five Y-Z
planes along the axial direction ……………………………………… 148
4.39 Velocity contours of three components in X, Y and Z directions at where x*
is of 0.25……………………………………………………………… 149
4.40 Four rakes of streamtraces are released at the upstream of the porous sample
at x* = 0.45 with contours of axial velocity. Each rake consists of 20
particles ocate at the heights of y* = 0.05, 0.25, 0.5 and 0.75 respectively
………………………………………………………………………... 149
B.1. 2D finite volume cells…………………………………………………... 164
xiv Acknowledgements
I am grateful to all those who have helped and encouraged me during my long journey in pursuing a Ph. D. degree. I am indebted to all of you. Your zeal for excellence and your personal warmth and kindness will continue to inspire me as I move forward to my next goals.
I must thank my advisor, Professor Isaac Greber for his guidance. I have learned a lot from him in both technical and non-technical discussions. I thank Drs. J. S. Tien, E.
B. White, R. V. Edwards and Mr. R. A. Blech for serving on my thesis committee, and for several helpful suggestions.
I wish to thank my husband, Meng-Sing and my boys, Yu-Ming and Deng-
Yuan. Without your support, I would not be able to go this far.
xv Nomenclature
A Channel cross-sectional area
Axi (Ax, Ay, Az), Cartesian components of the unit normal vector of
the cell face
B non-dimensional pressure drop based on the hydraulic diameter
Cf Inertial coefficient
Cv constant pressure specific heat for solid phase
c speed of sound
th ci molar concentration of the i species
D thermal diffusivity (K/(ρ·Cp))
D* dimensionless diffusivity (D/Din)
Dk numerical dissipation term
E vector of inviscid fluxes in the x-direction for gas-phase equations
2 2 2 Et total specific energy of gas phase, e + (u +v +w )/2
Ev vector of viscous fluxes in x-direction for gas-phase equations
F vector of inviscid fluxes in y-direction for gas-phase equations
Fv vector of viscous fluxes in y-direction for gas-phase equations
f non-dimensional effective heat conductivity of fluid phase in
porous system, [Kf]eff/ Kf
G vector of inviscid fluxes in z-direction for gas-phase equation
Gv vector of viscous fluxes in z-direction for gas-phase equations
h specific enthalpy of gas mixture
2 2 2 Ht total specific enthalpy of gas mixture, h + (u +v +w )/2
xvi Κ heat conductivity
K0 stagnant thermal conductivity of fluid phase
Keff effective thermal conductivity of fluid phase
Kd dispersion thermal conductivity
Kn Knudsen number, λ ⁄ l l local length scale of pore
L length of the numerical wind tunnel h height of porous samples ( height of channel as well) hsf interfacial heat transfer coefficient
n total number of cells
Nx number of mesh point in x-direction of the porous region
Ny number of mesh point in y-direction of the porous region
Ns number of gaseous chemical species
Nu Nusselt number (Equation 4.1)
p mean static pressure
Pe Péclet number (Pr·Re)
Pr Prandtl number (µCp/D)
qj heat flux in j-direction
qn net normal heat flux at the interface of solid and fluid cells
Re Reynolds number based on the hydraulic diameter (ρuh/µ)
Red Reynolds number based on the pore scale (ρul/µ)
0.5 Reκ Reynolds number based on the permeability (ρuκ / µ)
S source vector for gas-phase equation
xvii St Strouhal number
Tavg temperature averaged along the axial direction
T' fluctuation of temperature from the averaged temperature
T the temperature at the interface of solid and fluid cells
Tin inflow temperature
Th heated temperature at wall
Tm mean or volume averaged temperature (∫(ρuT )dA/ (ρVA))
* T invariant (dimensionless) temperature (Equation 4.2)
* U non-dimensional velocity magnitude with respect to uin
uin inflow velocity
V mean x-direction velocity = ((∫u dA/ A)
V' non-dimensional velocity vavg velocity in the y-direction averaged along the axial direction
(u',v', w') non-dimensional fluctuation components of velocity from the
respective averaged velocity components
th Vi volume taken by the i cell
Greek symbols
ρ mean density of gas mixture
µ viscosity of gas mixture
ε porosity
κ permeability (Equation 3.1)
τij viscous stress tensor component of Newtonian fluid
xviii λ mean free path
λmax the maximum eigenvalue of the preconditioning matrix
θ solid-to-fluid conductivity ratio ( Ks/Kf)
ω vorticity
∆n normal distance from cell center to face center
∆τ pseudo-time step
Subscripts f fluid s solid d dispersive eff effective w wall or surface quantity in inflow condition
Superscript
* dimensionless
xix A Numerical Study of Transport Phenomena in Porous Media
Abstract
by
May-Fun Liou
Since Darcy’s pioneering experimental study of porous medium flow, a great number of
analytical, numerical, and experimental works have been carried out to provide
qualitatively and quantitatively macroscopic descriptions of the overall viscous resistance
or heat transfer across the porous media. Recent advances in experimental measuring
techniques have uncovered the importance of structural heterogeneity within the porous
media. Thus, in order to gain a better understanding of phenomena at the scale of pores, new numerical approaches must be taken.
A general numerical simulation capability at pore-scale level is developed and validated in this thesis study, predicting global phenomena in close agreement with classical results. This technique has been successfully applied to two and three dimensional porous systems. In particular, it is shown that three dimensional solutions that couples the fluid and solid systems simultaneously at the pore scale are feasible with today’s computer resources and are extremely beneficial, shedding a new light into phenomena unavailable otherwise.
xx
This study also emphasizes numerical simulations of mass, momentum, and heat transfer
phenomena induced in complex porous media, providing details of local velocity profiles
and heat transfer. It is shown that the porous structures – shape, size, and locations have
significant effects on the macroscopic description. It is concluded that a microscopic description at the pore scale should be included in the study of porous medium flow. The flow pathlines are tortuous, determined by the local pore structure. Hence, the mixing caused by a porous insert can offer an efficient way to dissipate the heat from the heat source. The qualitative description of transport phenomena of flow through a three- dimensional duct demonstrates the capability of the numerical approach proposed in this thesis.
It is also found that the interplay among viscosity, heat conductivity and convection gives rise to a complex dynamical system. The effects of Reynolds number, Péclet number, local effective heat conductivity and properties of porous material on the local and global description of pressure, velocity field and heat transfer are studied in detail.
Finally, a summary of the thesis work and recommendation for future work are given.
xxi Chapter 1
Introduction
Natural and manufactured porous materials have broad applications in engineering
processes, including flow straighteners, heat sinks, mechanical energy absorbers,
catalytic reactors, heat exchangers, pneumatic silencer, high breaking capacity fuses, and
cores of nuclear reactors. For the subject of flow and heat transfer through porous media,
there have been extensive investigations covering broad ranges of applications since the early work of Darcy in the nineteenth century. Darcy correlated the pressure drop and flow velocity experimentally by defining a special constant property of the medium called permeability. However, it is only applicable to low speed (creeping) flow and low porosity saturated medium. It is well known that in flow through porous media the pressure drop caused by the frictional drag is proportional to the velocity at the low
Reynolds number range. In addition, this famous Darcy’s law also neglects the effects of solid boundary and the inertial forces on fluid flow and heat transfer.
Fluid transport is usually modeled using the continuum approach in terms of appropriate averaged parameters in which the real pore structure and the associated length scales are neglected. Moreover, those averaged parameters can only be obtained by experiments and are strongly influenced by the types of microstructure and operating conditions.
Fundamentally, they are limited to the scope of macroscopic phenomena. Specifically, the microscopic (pore scale) dispersion effect has significant impacts on the mass,
1 momentum, and thermal transports. Hence, modeling transport behavior at the pore-scale
for real engineering processes is desirable. In this study, an alternative numerical
approach is proposed and used for microscopic transport in porous media.
1.1 General Overview
Mass and thermal transport in porous media, such as ceramics, rocks, soils and catalytic
channels in fuel cells, play an important role in many engineering and geological
processes. There are two interesting aspects that arise in the research of porous media.
They are hydrodynamic and thermal effects. The dynamics of fluids flow through a
porous medium is a relatively old topic. Since the nineteenth century, Darcy’s law has
traditionally been used to obtain quantitative information on flow in porous medium. This
law is reliable when the representative Reynolds number is low whereas the viscous and
pressure forces are dominant. As the Reynolds number increases, deviation from Darcy’s
law grows due to the contribution of inertial terms to the momentum balance [Bear, 1972;
Kaviany, 1991]. It is shown that for all investigated media, the axial pressure drop is represented by the sum of two terms, one being linear in the velocity (viscous contribution) and the other being quadratic in velocity (inertial contributions). The inertial contribution is known as Forchheimer’s modification of the Darcy’s law
[Reynolds, 1900]. Basically, the pressure drop occurring in a porous medium is composed of two terms. Later Beavers and Sparrow [1969] proposed a similar model for
2 fibrous porous media. A general expression can be obtained from Bear [1972] and is widely accepted in the following formula,
dp µu =− (1.1) dx κ
It is seen that the pressure drop is directly proportional to the fluid viscosity µ and inversely proportional to the permeability of the porous mediumκ . Lage et al. [1997] suggested that an additional cubic term of fluid velocity be included in the above equation in the regime of higher speed (Reκ ≈ О(10)). Another significant work for predicting momentum transport in porous media is by Brinkman [1947]. Brinkman first introduced a term which superimposed the bulk and boundary effects together for flows with bounding walls. In Brinkman’s model, an effective viscosity was postulated from experiments performed on beds of spheres to replace the viscosity of fluid by taking into account of the porosity effect,
µeff = µ[1 +2.5(1 −ε)] (1.2) where ε is the porosity. There have been modifications on the above function to describe different types of porous media [Lundgren, 1972; Sahraoui, 1992]. More recently, computational modeling has been used to provide detailed flow fields. There are also results obtained by the asymptotic solutions [Chapman & Higdon, 1992].
While study of porous media flow is an old topic in fluid mechanics, the convective heat transfer of flows through porous medium has emerged as a new interest due to new technologies developments. Forced convection in porous media arises wherever the energy is delivered, controlled, utilized, converted or produced. The recent widely used
3 cellular microstructure materials have found implementation in the technologies of thermal dissipation media, impact absorbers and compact heat exchangers. Their thermal attributes enable applications as heat dissipation media and as recuperation elements.
Consequently, these enable high heat transfer rates and can be effectively used for either cooling or efficient heat exchange. Hence, it has become important to understand the interaction between mass and thermal transports and the resulting effects on the thermomechanical characteristics of porous media.
There has been considerable analytical, numerical, and experimental work done in the past to measure or estimate the overall heat transfer rate of convective heat transfer in porous media. Some early experimental studies have investigated the flow of fluids through packs of spheres [Lage et al., 1997; Kocecioglu & Jiang, 1994] or real porous material samples [Boomsma & Poulikakos, 2002]. Also, resurging interests in applying complex micro-structures to chemical industries, transpiration cooling etc., have prompted studies in experiments of highly porous media [Hunt & Tien, 1988a]. In general, the experimental results usually are described (postulated) statistically as empirical relationships in terms of dimensionless parameters of permeability, Reynolds,
Nusselt, Prandtl, and Peclet numbers. However, those correlations fail to address general mass and thermal transfer phenomena other than specific configurations, such as packed bed of particles, or highly porous open-cellular foams.
Most recently, non-intrusive experimental techniques have been applied to allow some detailed visualizations of microscopic flow and mass transfer occurring within porous
4 media. Among these are Laser Doppler Velocimetry (LDV), Particle Image Velocimetry
(PIV), and photoluminescent volumetric imaging (PVI). Dybbs and Edwards used a laser
anemometer for measurement inside model porous media. The measurement of local
velocity fields above the porous medium was first attempted by Vignes-Adler et al.
[1987], using LDV and Stephenson and Steward [1986] using PIV. Since then the experimental measurement techniques and devices have been advanced from differential pressure transducer and hotwires in packed bed of glass beads or cylinders [Beavers et al., 1974; Vafai, Alkire & Tien, 1985] to LDV and 100µm thermocouples [Okuyama &
Abe, 2000], or Magnetic Resonance Imaging (MRI) [Shattuck et al., 1991, 1995; Li et al.,
1994; Manz et al., 1999]. Notably, Saleh, Thovert and Adler [1993] used PIV to measure velocity close to porous media and pointed out the deficiency of the classical semi- empirical Brinkman equation [Brinkman, 1947]. However, these optical techniques can only be applied to transparent material and solid materials with optical index in the same range as the fluids. Meanwhile, Magnetic Resonance Imaging has been adopted for nonmagnetic materials to record the interstitial velocity distributions [Kutsovsky, et al.,
1996; Ogawa et al., 2001; Suekane, et al., 2003] for Darcian flow and non-Darcian flow respectively in packed beads and crushed glass samples. The obtained results for Darcian flow indicated a strong non-uniformity in velocity. In places, reversed flow is induced by pore structure. In particular, Sederman et al. [1998] observed a significant heterogeneity in the flow with 8% pores carrying 40% of the flow volume. Also, they found that high- volume flow rate is influenced mostly by the morphography of pores. These detailed measurements on the microscopic level indicate that mass and heat transfer in a packed bed of beads or crushed glass are sensitive to the interstitial velocity distributions in pores
5 over a wide range of Reynolds numbers (1 ≤ Red ≥ 210). These studies all point out that the structural heterogeneity of the porous region plays a critical role in the transport of mass and heat inside porous media.
Concerning analytical studies, Darcy’s law ignores the effects of solid boundary or the inertial forces on fluid flow and heat transfer. While these effects become significant near the boundary and in highly porous materials, relatively little attention had been directed to study these effects. Notably, Brinkman [1947] proposed a model to account for the presence of solid boundary by adding a viscous term to the Darcy’s law. Muskat [1946] took inertial effect into account by adding a velocity squared term to the Darcy’s law.
However, both works do not consider boundary and inertia effects simultaneously. Vafai and Tien [1981] used a volume-averaged momentum and energy equations to numerically study both effects simultaneously.
The volume averaging method currently is widely used for investigating inertial and boundary effects on flow and heat transfer in porous media. It uses local volume- averaging technique [Slattery, 1967 & 1970] and supplementary empirical relations to establish the macroscopic governing equations of momentum and energy for porous media [Vafai & Tien, 1981; Vafai et al., 1985; Calmidi & Mahajan, 1999]. These empirical parameters include the inertial coefficient Cf in the momentum equation and the
effective thermal conductivity Keff in the energy equation, which consists of stagnant K0 and dispersion thermal conductivity Kd. Conventionally, the stagnant thermal conductivity is determined from porosity and the thermal conductivity of the solid and
6 the fluid phase. Moreover, some models consider extra terms of interfacial heat transfer to account for the diffusion between two phases [Wakao, 1979]. An empirical model for its determination was presented by Tien and Vafai [1979]. In more advanced approaches, a geometric averaging technique is used for estimating heat transfer [Hsu, 1994 & 1995].
Also, Calmidi & Mahajan [1999] and Boomsma & Poulikakos [2001] developed models by utilizing geometrical estimate for calculating the effective thermal conductivity, specifically for metallic foams saturated with fluid. Calmidi & Mahajan proposed a one dimensional heat conduction model of a two dimensional array of hexagonal cells representing the porous medium. Later Bhattacharya et al. [2002] extended the Calmidi &
Mahajan model with a circular intersection, which results in a six fold rotational symmetry. Also Boomsma and Poulikakos used tetrakaidecahedron cells with cubic nodes at the intersection of two nodes to model highly porous metal foams. Both geometrical models are based on experimental data to provide the geometric parameters.
Recent applications of Lattice-Boltzmann methods provide pore-scale information and give promising results [Martys & Hagedorn, 2002; Manz & Gladden, 1999; Verger &
Ladd, 1999; Szymczak & Ladd, 2004]. The major difficulties in those approaches are how to model flows in heterogeneous materials and how to properly handle the interfacial boundary conditions between fluid and solid matrix at the pore scale [Bauer, 1993;
Thompson & Fogler; 1997; Thompson, 2002].
Though the local volume-averaging method is a more practical approach than the analytical counterpart, it does not provide a general approach in describing transport in
7 porous media. Since the volume averaging is still in use today and is the framework for
studies of porous media, the next section will be devoted to the volume-averaging
method.
1.2 Volume Averaging Methods
y+∆y
up y
x v+(∂v/∂y)∆y x+∆x z+∆z y+∆y
z w+(∂w/∂z) ∆z
u+(∂u/∂x) ∆x u
w
y x Representative x+∆x Elementary v Volume (REV)
Figure 1.1: Schematic of porous medium, flow variations in the Representative
Elementary Volume and the pore velocity vector u p .
The local averaging performs the averaging of the microscopic equations over a
representative elementary volume (REV) as shown in Figure 1.1. Note here, the REV is
8 the smallest volume possessing local averaged quantities that are statically meaningful.
Its definition and the relevant mathematical operations for partial differential equation
(e.g. N-S equations) may be found in Slattery [1967] and Kaviany [1991]. A scalar
quantity φ of a fluid is averaged over a fluid volume Vf ,
1 φφ =dv (1.3) V ∫ vf
Let the porosity of a porous medium ε = Vf / V, then
1 ⎛⎞1 φφ = dv = ε ⎜φdv⎟ (1.4) VV∫ ⎜⎟∫ vvff⎝⎠